Using Manufacturers' Specifications as Type B Error Contributions

About this Article
This article by Robert L Brown was presented at the NCSLI 2006 Symposium and published in the Proceedings.

Abstract

Manufacturers' specifications are a complicated interplay of consumer demand, contractual agreement and definition of "fitness for use" warranty. In order to better understand the implications of using a manufacturer's specification in an uncertainty analysis, we will need to explore technical topics such as the following...

• How are specifications created and managed?
• Advantage of using a specification.
• Statistics verses managed specification.
• Stationary and non-stationary random processes.
• GUM concepts like "safe".
• Issue and definition of pseudo systematic error.

This paper examines these issues in the context of designing a calibration standard. As a result, it will become clear why an uncertainty analysis (employing Type B data) is at best a worst case analysis. This can effect how calibration laboratories use uncertainty data in the quality system and on customer facing documents or training.

The simulation mentioned in the paper which demonstrates the Central Limit Theorem can be viewed by clicking the following link. And don't worry, this is one gambling machine that you won't lose money playing!

• Pachinko Gambling Machine Simulation

Note that your computer must be able to run a Java applet to view the simulation.

View or Download

The entire paper is available to view or download in Portable Document Format. But if your browser doesn't already have the (free) viewer installed, you will first need to get and install the Acrobat® Reader™.

	Manufacturers' Specifications as Type B Error Contributions
	PowerPoint slide show